Fig 2: SAFB1 colocalizes with calbindin and SMI‐34 in the cytoplasm of SCA Purkinje cells. A. Calbindin immunopositivity specifies Purkinje cell soma and dendrites. SAFB1 staining is absent from the cytoplasm of control Purkinje cells but is colocalized with calbindin in SCA cases (arrows). Cell nuclei were stained using DAPI (blue). B. SMI‐34 staining, a marker of Purkinje cell damage colocalizes with SAFB1 in the soma and dendrites of SCA cerebellum (arrows) but not control tissue. SMI‐34 also stains basket cell projections (asterisk), which appear hypertrophied in SCA tissue compared to control. Scale bar is 50 µm.

Fig 3: Increased binding of SAFB1 to Ataxin 1 with an expanded glutamine tract. A. The top 10 most SAFB1 bound RNA transcripts with the total number of iCLIP SAFB1 cross‐link sites from three independent experiments. B. The red line depicts the location of the ATXN1 gene on chromosome 6. The majority of the SAFB1 binding sites occur in RNA encoded by exon 8 of ATXN1 and in the region containing CAG repeat sequence (blue hatched lines). A screengrab taken from the genome browser shows the sense (brown) and antisense (blue) SAFB1 iCLIP binding sites. The binding sites relative to the amino‐acids encoded by the SAFB1 bound sequences are shown in a 3′‐5′ orientation. Note the SAFB1 cross‐link site will correspond to the amino‐acid most 5′ to the sense sequence and 3′ to the antisense sequence. The numbers within the schematic of the sense binding sites show the number of SAFB1 cross‐link sites in each of these individual experiments. C i. Western blotting of protein lysates from RNA immunoprecipitation experiments, probed for SAFB1. SAFB1 is present in both ATXN1‐30Q transfected and ATXN1‐85Q transfected cell lysates immunoprecipitated using a SAFB1 antibody (IP). There is no SAFB1 protein detected in IgG control lysates (IgG). Expected molecular weight for SAFB1 is 102KDa (arrow). ii. Relative levels of ATXN1 RNA bound to SAFB1 protein. SAFB1 protein was immunoprecipitated from HeLa cells transfected with plasmids containing ATXN1‐30CAG (physiological) or ATXN1‐85CAG (pathological expansion). The relative expression level of ATXN1 RNA associated with SAFB1 protein is significantly increased in ATXN1‐85 CAG compared to ATXN1‐30 CAG (unpaired, two‐tailed t‐test t(4) = 3.76, P = 0.01). Data are presented as mean ± SEM.

Fig 4: Frequency and localization of SAFB1 staining compared with polyglutamine inclusions in HD. A i. No primary antibody control staining of striatal tissue from an HD patient (upper panel). Polyglutamine immunostaining using antibody 1C2 in the striatum of a control patient shows no immunoreactivity (lower panel). ii. Representative low‐magnification images of polyglutamine and SAFB1 staining in separate, serial striatal sections of an HD patient (case 23). The percentage of neurons positive for cytoplasmic SAFB1 and for polyQ is indicated below the images. B. High magnification images show that polyQ aggregates/inclusions occur in both the nucleus and cytoplasm. SAFB1 staining is generally homogenous, with strong staining throughout the nucleus and an even distribution of staining with small puncta present throughout the cytoplasm. Scale bar in (A) is 50 µm, in (B) is 10 µm.

Fig 5: SAFB1 regulated dendritic spine size in primary hippocampal neurons. a Immunocytochemical analyses show SAFB1 is highly expressed in the hippocampus (i). SAFB1 is clearly evident in pyramidal neuron cell bodies in the CA1 region of the hippocampus ((i) red SAFB1 immunofluorescence) and a neurofilament stain of dendrites (green immunofluorescence (ii) second panel) shows SAFB1 is located exclusively in neuronal cell bodies. Primary hippocampal neurons were transduced with SAFB-EGFP or EGFP alone (confocal images shown in (b) and confocal microscopy used to measure spine (c) area, (d) size, and (e) density. Spine size measures were analysed by cumulative frequency using the Kolmogorov-Smirnov test (b, d = 0.1075; P <0.001) and by t-test (c) ***P <0.001. Values shown in bar graphs are means ± SD